Grant Muller, Presenter: Nothing to Disclose

Peter Karasev, Abstract Co-Author: Nothing to Disclose

Ivan Kolesov, Abstract Co-Author: Nothing to Disclose

John Xerogeanes, Abstract Co-Author: Nothing to Disclose

Allen Tannenbaum PHD, Abstract Co-Author: Nothing to Disclose

A sufficiently large dataset and novel software for precise MRI shape extraction enable a detailed understanding of bone structure change over time. An MRI scan of a new patient can use the pattern to for prediction of growth potential and surgical planning. Ionizing X-ray exposure can be avoided in the computation of bone age using an accurate MRI shape-based method. 

Bone growth and development in pediatric knee patients will be studied over a significant population, with an emphasis on establishing methodology for the identification of bone age from extracted 3D models. Recently developed algorithms and software facilitates the extraction of detailed shape volumes of the physis and distal epiphysis from MRI scans of adolescents patients experiencing normal bone growth. Segmentation is the task of delineating an anatomical structure from the surrounding tissue. Statistical shape analysis can be applied to the segmented models, thus quantifying structural variation and its relationship to bone age. It is hypothesized that bone age can be accurately estimated based on analysis of the extracted physis and distal epiphysis shapes. 

Approximately 400 anonymous high resolution T1 coronal MRIs of the knee have been gathered, with known patient age. These scans are of healthy adolescents between the ages 3-18; for each age group, ten male and ten female MRIs will be used, in total approximately 320 scans. For each dataset, software developed specifically for this application is used to interactively segment the organs of interest: epiphysis, physis, metaphysis and cartilage cap of the distal femur and proximal tibia. From this segmentation, Once all models are complete, statistical analysis is performed within each age group to establish a relationship between parameters describing the models' shapes and the corresponding age of the patient. 

The software required to perform the segmentations was recently completed. It has been successfully used to create 3D models of the bones for the majority of our MRI scans. In parallel, the models for the completed age groups are being analyzed to determine their shapes' correlation to patient age.

Muller, G, Karasev, P, Kolesov, I, Xerogeanes, J, Tannenbaum, A, Physis and Epiphysis Structure Analysis from MRI for Bone Age Prediction and Surgical Planning.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11006720.html